ECOS4016F: Econometrics | My Assignment Tutor

University of Cape TownSchool of EconomicsECOS4016F: Econometrics 2021Time series assignmentAdmin• The assignment will be done individually.• Final Hand in: Friday 14 June at 10:00.• Word limit: 3000 – 3500 words (after 3850 words, penalties will apply)• Formatting requirements: 1.5 spacing, font size 11, 2cm page marginsDescriptionFor this assignment, you are tasked with developing a forecast of quarterly gross domesticproduct (in constant prices) in South Africa. Your forecast should include a point estimateas well as 95% confidence intervals for the 3 year period from January 2018 through to December 2020. In developing a forecast for the period mentioned, you will be able to assessthe extent to which Covid-19 affected the expected trajectory of GDP by comparing yourforecasted values with the realized GDP.You will need to evaluate a number of different regression specifications and decide on 3models which you intend to test. You will need to decide which variables to include in thesemodels and motivate why these variables may be important.You will be expected to perform all of the necessary tests on the data, including testing forheteroskedasticity, stationarity and serial correlation. You will then be required to explainand motivate how you intend to deal with these issues, the trade-offs you face and how thesewill affect your coefficient estimates, standard errors and goodness of fit measures.The assignment is intended to test the extent to which you are able to practically apply theskills you have learnt throughout this course. The assignment should take the form of a miniacademic paper. As such, you must use academic writing. No bullet points, and make sureyou use proper sentences, grammar, punctuation and referencing.DataFor this assignment, you will be required to download macroeconomic data for South Africafrom the South African Reserve Bank (SARB) Quarterly Bulletin. All data can be downloaded via the SARB’s statistical query service, here. You can find a description of the available variables and their codes, here. You may use any additional macroeconomic datasetsyou think will be relevant, so long as these datasets are properly cited.1Assignment structureYou assignment will have 6 sections, namely1. IntroductionIntroduce your research question, the data you are using, the models you will be using andpreview the results.2. Data and descriptive statisticsIn this section you are to introduce the data you are using, the time period it covers andthe source. You must produce a descriptive statistics table containing information on: thenumber of observations; mean; standard deviation; minimum and maximum; P25, P50 andP75. If in later sections of the paper you perform any transformations on the data, you areto also include the transformed variables in this table. You are also expected to producefigures showing the evolution of key variables. Thereafter, discuss any notable features ofthe data.3. DiagnosticsYou will be expected to perform and report any tests you conduct on the data, includingtesting for heteroskedasticity, stationarity, serial correlation. You will also be required todiscuss the extent to which the OLS unbiasedness and consistency assumptions are likely tohold. You will then be required to explain and motivate how you intend to deal with theseissues (transformations, adjustments to standard errors, time trends, alternative estimationtechniques etc.) the trade-offs you face and how the decisions you take will affect yourcoefficient estimates, standard errors and goodness of fit measures.4. Model specificationAfter completing the diagnostics, you are required to specify 3 regression models for yourforecast.1 You need to motivate the choice of specification as well as motivate any covariatesyou decide to add and why you have decided to add them. If you need to perform additionaldiagnostics at this point to arrive at your choice of model, you should report the results.5. ResultsIn this section you are to report the regression output from your 3 models and discuss.Thereafter, you need to evaluate the accuracy of the forecasts, using the RMSE and/or theMAE. Once you have determined the best fitting model, you need to present the forecastfrom this model graphically. This should take the form of a line graph, which plots the data,1Once you complete this step, you may need to amend your descriptive statistics and diagnostics. Forexample, if in this step you decide to add a covariate to your model in the anticipation that the covariatewill be an important regressor, you will need to make sure that you adequately discuss that regressor in theprevious sections.2together with the forecast and confidence intervals. The forecasted line should be clearlydistinguishable from the data line and the confidence intervals should be in a different color.6. DiscussionOnce you have identified the best model, discuss how the forecasted path of GDP differedfrom realized GDP and how much lower realized GDP was from forecasted GDP, discussingin particular the role of Covid-19. You can motivate your discussion using evidence fromother sources, including for example, employment data from the Quarterly Labour ForceSurvey produced by Statistics South Africa and/or NIDS-CRAM.7. ConclusionBriefly summarize your findings.CodeThe assignment must be completed in R, using RStudio. You will be required to submit yourR script and your code will be evaluated on the extent to which it adheres to good codingpractice. See the assignment rubric for more details.3


Leave a Reply

Your email address will not be published. Required fields are marked *